[GHC] #9221: (super!) linear slowdown of parallel builds on 40 core machine

[GHC]

#9221: (super!) linear slowdown of parallel builds on 40 core machine
-------------------------------------+-------------------------------------
        Reporter:  carter            |                Owner:
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:  8.2.1
       Component:  Compiler          |              Version:  7.8.2
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #910, #8224       |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by rrnewton):

 Would people agree that we should not have spinlocks in the system without
 some back-off strategy?  E.g. switch to sleeping/blocking after some
 number of iterations.

 Even a constant upper bound could reign in worst case behaviors.
 Apparently OS's provide some notion of "adaptive lock" that has other nice
 features, e.g.:  http://stackoverflow.com/questions/19863734/what-is-
 pthread-mutex-adaptive-np

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9221#comment:62>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler